Tony's blog

Data Domain - Built for Inline Dedupe

It is important to consider Data Domain's architectural approach optimized for inline data
dedupe. 

Data dedupe uses algorithms to analyze and make changes to data with the goal of reducing the amount of capacity to store it. Addiitonally, a core tenant of data dedupe is to do no harm - all of the data should be accessible and unaltered from the user and application perspective.  These characteristcs make inline data dedupe CPU intensive.  And for Data Domain customers - that is a very good thing.

Traditional storage system performance is typically disk-limited with disk drives as the biggest bottleneck.  Many storage systems offset this by providing caching algorithms and wide stripping across a large number of disk drives.  Wide stripping is valuable but again relies on disk performance - which is the slowest aspect of any storage system.  The concept behind wide stripping is "many hands makes light work".  But those "hands" or disk drives don't come for free - additional disk capacity results in high costs to the customers on a number of levels. 

There is brilliance in the CPU versus disk-driven approach for a data dedupe storage system.  There is a wide gap between CPU and disk performance and this continues to increase over time.  Additionally, the cost of CPU performance is far lower than disk drives.  And since data dedupe - especially when it is being performed inline - is CPU-intensive - Data Domain has a sustainable architectural advantage. 

Traditional storage systems were built to support transactional data and not to be dedupe engines.  That is why you see very few inline data dedupe storage systems in the market.  Most storage system vendors have a "hammer" and they want to turn everything into a "nail". 

Data Domain's architecture enables their ability to provide inline data dedupe with high performance levels that will just keep getting better as CPU performance keeps getting better.  This is critically important since the growth of primary data increases and all of it has to be backed up without impacting operations or backup windows. 

Data Domain isn't a general purpose storage system with data dedupe algorithms added as a feature - it is a purpose-built solution that was designed and optimized for inline data dedupe that has created an entirely new and valuable category in IT. 

Setting the Record Straight

The CEO of EMC, Joe Tucci, recently claimed that they are the number one vendor in data dedupe. Maybe the statement would be more defensible if he didn't state that EMC was beating Data Domain in the market. This is not the case.

The first issue is that much of the revenue EMC derives from data dedupe is from their Avamar backup software - which is not a competitive solution to Data Domain. If they want to do an apples-to-apples comparison they need to compare their disk-to-disk (D2D) backup appliances - the EDL with data dedupe directly to Data Domain - and that number is nowhere close. Unfortunately EMC can say whatever they want - because they don't have to prove it. On the other hand - Data Domain's success is transparent and open for the world to scrutinize.

Why not add Centera revenue into the mix as well since it does single instancing - a less efficient form of data dedupe? That would really boost their numbers against Data Domain. After all, they add Centera to their NAS numbers and claim the number one spot over NetApp - and no one takes that seriously either.

The reason that this is important is because just lke NetApp established their leadership in NAS - Data Domain is doing the same for D2D dedupe backup and their competitors will create FUD to try and slow this down. So it is important to provide anti-FUD every now and then.

Data Domain is the leader in D2D backup data dedupe appliances - with the most mature and advanced products, the greatest number of implementations, the largest number of customers, the smartest and most experienced support organization, an extensive ecosystem of partners and resellers, the highest revenue numbers and a very strong business model with great margins and real momentum in the market.

Customer Discussion: Bank Midwest

I had another great discussion with a Data Domain customer- Bryan Wilken, VP of IT for Bank Midwest:

Tony: Can you give us any examples of how Data Domain provided real value to your company?

Bryan Wilken:  e have completely stopped using tapes in all but one of our systems, and we’ll have that on the Data Domain in the next few weeks here. The bank now has a more robust and fault-tolerant back-up system. It runs automatically every night and it doesn’t require the need of employees to change tapes and take them home. We’re also seeing a 23-to-1 compression ratio, or de-duplication. And restores have been handled quickly and effectively by the IT department. We no longer have to wait to have tapes either brought back in or reloaded and reread before we can restore data.

Tony:  Can you tell us how Data Domain has improved the overall economic effectiveness within your IT operations?

Bryan Wilken: The Data Domain disk-to-disk system is saving Bank Midwest in tapes and drives alone around $13,000 a year. Plus we have the security risk that’s been eliminated without having tapes that are leaving the bank’s premises. As I mentioned, we have moved to virtualization of our servers. We hadn’t first planned on this, but we are now backing up our VMware servers nightly to our Data Domain and then replicating them to our DR site. Another added layer of protection in case of a disaster.

Tony: Would you recommend using Data Domain to other IT professionals?  And tell us why.

Bryan Wilken: Yes, I would. What I liked about Data Domain as we’ve gone through installing this product, is that the solution worked just as the vendor claimed it would.  We’re also getting the compression rates that they claim and it was just an easy set-up, and it’s worked from the day it was installed. That’s the kind of products I like to purchase for Bank Midwest.

Bryan validates the Data Domain value proposition with 23-to-1 data dedupe, fast and reliable recoveries, reduced management overhead of tapes, reduced security risks, enabled disaster recovery - and fundamentally it was easy to set up and put into production. 

I think it is interesting to note that this is yet another customer that is completely eliminating tape.